RPython is the language used to create PyPy. Technically, it is considered a translation and support framework for implementing dynamic programming languages, separating the language specs from the implementation aspects. This means that RPython can be used for other languages besides Python, though it is most commonly associated with Python. This also means that any dynamic language will benefit from the JIT compiler and allows for a mix-and-match style when making implementation choices.

While certain environments have been created in the past to provide abstraction between source code and the target system, such as .NET and Java Virtual Machines, RPython uses a subset of CPython to create languages that act as simple interpreters, with little direct connectivity to low-level, system details. The subsequent toolchain creates a solid virtual machine for a designated platform by using the appropriate lower-level aspects as needed. This allows further customization of features and platform configuration.

When implementing languages, developers have to contend with the languages themselves, the platforms that will run the languages, and the design decisions made during development. The overriding goal of PyPy and RPython development is to make it so that these development variables can be modified independently. Thus, the language used can be modified or replaced, the platform-specific code can be optimized to meet different model needs and desired trade-offs, and the translator backends can be written to target different physical and virtual platforms.

Thus, while a framework such as .NET tries to create a common environment for developers to target, PyPy strives to allow developers to essentially do whatever they want, however they want. JIT compilers are one way to do this, as they are made in a language-independent manner.